In certain situations involving cloud computing, data may need to be transferred from a customer data source system to a target on a cloud system. When loading data from an on-premise source to a target on a cloud system, the network connection can be dropped at any time, causing the current data loading job to fail. If the job has to transfer millions of records from the source to the target and the connection is compromised, for example, during the last batch of records, then all of the records need to be transferred again from the source to the target. This method of data loading is inefficient.